Japanese encephalitis vaccines: Immunogenicity, protective efficacy, effectiveness, and impact on the burden of disease

ABSTRACT

Japanese encephalitis (JE) is a serious public health concern in most of Asia. The disease is caused by JE virus (JEV), a flavivirus transmitted by Culex mosquitoes. Several vaccines have been developed to control JE in endemic areas as well as to protect travelers and military personnel who visit or are commissioned from non-endemic to endemic areas. The vaccines include inactivated vaccines produced in mouse brain or cell cultures, live attenuated vaccines, and a chimeric vaccine based on the live attenuated yellow fever virus 17D vaccine strain. All the marketed vaccines belong to the JEV genotype III, but have been shown to be efficacious against other genotypes and strains, with varying degrees of cross-neutralization, albeit at levels deemed to be protective. The protective responses have been shown to last three or more years, depending on the type of vaccine and the number of doses. This review presents a brief account of the different JE vaccines, their immunogenicity and protective ability, and the impact of JE vaccines in reducing the burden of disease in endemic countries.

KEYWORDS:

• duration of immunity
• genotypes
• Japanese encephalitis
• vaccines
• vaccine benefit
• vaccine effectiveness

Japanese encephalitis epidemiology, etiology, and the need for vaccines

Japanese encephalitis (JE) was first described in 1871,¹ and frequent outbreaks have since been recorded in more than 20 countries in South Asia and the Western Pacific. In fact, JE is the major viral encephalitis in this region, and has been described as ‘a plague of the Orient.’² About 43% of the human population, about a quarter of whom are below the age of 15 years, lives in areas at risk of JE.³ The disease is caused by the JE virus (JEV), a flavivirus transmitted by mosquitoes. The distribution of the disease overlaps with that of the most common JEV transmission-competent Culex species, particularly C. tritaeniorhynchus and C. vishnuii; the virus can also infect other mosquito species,^4,5 raising the possibility of wider transmission competency. The disease is an occupational hazard; people working in rice fields and other water-logged areas, as well as owning pigs or living near pig populations, are more at risk to be infected with JEV and develop the disease.^6,7 Wading ardeid birds, especially herons, serve as reservoirs for JEV, and pigs serve as amplifying hosts, whereas humans, cattle and horses serve as dead-end hosts.^8-12

The majority of JEV infections are asymptomatic, with only about 0.2–0.4% of the infected individuals developing encephalitis. Approximately 67,900 cases of clinical JE are estimated to occur every year, about 75% of them being in those under 15 years of age.¹³ While these assessments may be underestimates due to the poor diagnostic and surveillance capabilities in many countries, extensive serological cross-reactivity between various flaviviruses also presents problems in definitive determination of the etiology. The disease, which follows an incubation period of 5–15 days, is characterized by initial symptoms such as fever, inappetence, headache and malaise. About 20–30% of those who exhibit these generalized symptoms develop neurological signs, including nuchal rigidity, convulsions, and hemiparesis. The disease is fatal in 20–25% of symptomatic subjects, and ∼40–45% of the recovered individuals suffer from long-term neuropsychiatric sequelae such as hemiparesis, deafness, disorientation, and mental instability or retardation, as well as general neurological symptoms such as headache, nausea, and vomiting, and sometimes, orchitis.¹⁴

Japanese encephalitis virus (JEV) was first isolated in 1924. It is an enveloped virus containing a single-stranded positive sense RNA genome, which is organized in the order representing the proteins C-prM-E-NS1-NS2A-NS2B-NS3-NS4A-NS4B-NS5 (C: capsid; prM: pre-membrane; E: envelope; NS: non-structural). The genome is translated into a single polyprotein, which is cleaved both by cellular and viral proteases, resulting in three structural (C, prM, and E) and eight NS proteins, including the NS1′ protein which is produced via ribosomal frameshifting as a consequence of an RNA pseudoknot structure.¹⁵ The virion is formed by the C protein forming an icosahedral cage encasing the genome, and being enveloped by host cell membrane studded with the E protein.¹⁶

The JEV exists as five distinguishable genotypes – G-I, G-II, G-III, G-IV and G-V – based on nucleotide homology in the E protein gene; three subtypes (a, b and c) are also distinguishable within G-I.^17,18 Besides the genotypes, several different subgroups and haplotypes have also been reported.^19,20 However, all JEV strains belong to a single serotype, as evident from (a) epidemiological observations of the absence of secondary encephalitis,²¹ (b) the fact that vaccination has significantly reduced JE cases (see below), and (c) studies of in vitro neutralization of heterologous genotype viruses by sera from subjects immunized with any one of the genotype viruses.^17,22-29 Types G-I to G-IV have been frequently detected whereas G-V is relatively rare.³⁰ Viruses belonging to G-I have been reported from Australia, Cambodia, China, India, Japan, Malaysia, South Korea, northern parts of Thailand, and Vietnam; G-II has been reported from Australia, Indonesia, Malaysia (especially Sarawak), Papua New Guinea, South Korea, and southern parts of Thailand; G-III has been reported from China, India, Japan, Nepal, The Philippines, South Korea, Sri Lanka, and Taiwan; G-IV has been reported from Indonesia; G-V has been reported from China, Malaysia, and South Korea^31,32 It should be noted that in most endemic countries, more than one genotype may circulate simultaneously, with possible periodic changes in dominant genotypes.³³ Only a handful of isolates of G-V have been reported, that too from mosquitoes and pigs,^30,3435 and it has been difficult to isolate G-V viruses from humans. However, G-V virus(es) has/have been isolated from at least four different species of Culex mosquitoes during and after a recent outbreak in South Korea.^36,37 Some association of the circulation of the genotypes with climate and geography has been proposed,^38,39 although more work is required in this direction as well as in the area of the competency of the various vector species to transmit the different genotype viruses.

Up until recently, G-III was the most prevalent JEV genotype, but it has been replaced by G-I in many countries.^20,40,41 Despite cross-neutralization studies with different genotype viruses strongly suggesting cross-protection in vivo, the reasons for the shift from G-III to G-I are unclear, although molecular analyses have suggested a role of restriction in host range and replication proficiency.⁴¹ However, no detailed experimental analysis comparing various strains or serum samples have been performed, and the lack of reference reagents (genotype viruses as well as panel of antisera against each genotype virus) is an impediment for carrying out such comparative analyses (see below). It is important to note that serological cross-reactivity and/or cross-neutralization alone may not be sufficient, and a quantitative cut-off level may be needed to define serological subtypes among the various strains. Indeed, G-V viruses are somewhat divergent both at the nucleotide and at the amino acid level (21 residues being different from those of a G-III virus),⁴² and have recently been shown to be poorly neutralized by sera against other genotypes,⁴³ stressing the importance of carrying out extensive serological reactivity and antigenic mapping analyses. In this context, it has recently been shown with dengue virus that divergence at antigenic level among viruses within a serotype was not significantly different than among viruses belonging to different serotypes.⁴⁴ It is therefore possible that minor antigenic variation could partly influence super-infection with different genotypes or subtypes of JEV, or the severity and/or outcome of JE disease.

There is no specific treatment for JE, and the only recommendations are for palliative therapy. Besides the regionally widespread nature of the disease, and the threat to travelers and armed forces personnel commissioned in endemic areas, there is always a potential for JE to spread beyond its traditional borders, because of the possibility of movement of infected animals, birds or mosquitoes. Control of mosquitoes through chemical or biological means are an option, but such strategies could potentially lead to ecological imbalance since they could affect all the mosquitoes. Another way is to specifically target the reduction of Culex mosquitoes by genetic engineering, but this has not been tested widely in the field, not to mention potential ecological consequences. On the other hand, it is near impossible to control JE in pigs due to the unorganized nature of pig-keeping in several of the endemic countries, their prolific reproduction rates and the wallowing habit of pigs in mosquito-breeding habitats. Following the onset of rains, a high proportion of pigs get infected with JEV and produce high level of viremia,¹⁰¹² and are essentially the first indicators of circulating JEV in an area before outbreaks in humans. Given these constraints, the best way to control JE in humans is through vaccination.

Immunological determinants of protection against Japanese encephalitis

Neutralizing antibodies (NAbs) at 50% plaque reduction neutralization titer (PRNT₅₀) of at least 10 have been established as a correlate of protection against development of JE disease in humans.,^21,45,46 In a well-designed study where human sera with no, low or high NAb titres were tested in protecting mice against challenge with virulent JEV strains, it was shown that a titer of ≥10 was protective, although partial protection could be achieved with titres of <10.⁴⁶ For vaccine end-point analyses, the PRNT₅₀ values are converted into seroprotection rate (SPR; proportion of subjects with titres of >10) seroconversion rate (SCR; proportion of subjects having baseline and immune titres of <10 and >10, respectively, or >10 and at least a 4-fold increase, respectively), and geometric mean titres (GMT). However, there is no standardized protocol for neutralization test; various cells [chick embryo fibroblast (CEF), primary hamster kidney (PHK), Vero, baby hamster kidney-21 (BHK-21), LLC-MK₂], different strains of virus, different visualization techniques (plaque formation, immunohistochemistry, focus formation), and different cut-off values (50% versus 90% PRNT) have been used. An attempt was made to standardize the PRNT assay from reagents and protocols followed in 11 different assays by five different participants, but proved unsuccessful, mainly due to the differences in challenge virus strains used for neutralization,³¹ suggesting that reference materials may need to include multiple virus strains as well as sera for a meaningful comparison.

In contrast to recognition of antibodies as markers of immunogenicity and protection, very little is known about the role of cell-mediated immunity (CMI) in protection against JE in humans. T-lymphocyte responses were shown to be elicited in individuals immunized with recombinant pox viruses expressing prM, E, NS1 and/or NS3 proteins of JEV.^47-49 At least some of the CD4⁺ T-lymphocyte clones were specific to E protein and have been shown to exhibit lytic function in vitro,⁴⁸ suggesting that this might be important for virus clearance in vivo. A systematic study found that in natural infection, the dominant lymphoproliferative as well as interferon (IFN)-γ response was directed to NS3, and that the response was poor against E, NS1 or NS5; this study also found preferential activation of CD4⁺ over CD8⁺ T cell responses against the E protein, and that the opposite was observed for NS3, whereas there was no such bias with NS1 or NS5.⁵⁰ A recent study found that healthy JEV-exposed individuals elicit a predominant CD8⁺ T cell response which is directed toward NS3, NS4 and NS5 proteins, and that elicitation of a polyfunctional CD4⁺ T cell response was a positive prognostic indicator of disease outcome.⁵¹ The CD4⁺ T lymphocyte responses are likely to play an important role in long-term anamnestic humoral responses, although this has not been systematically studied.

Japanese encephalitis vaccines

A number of JE vaccines have been developed, and these include inactivated and live vaccines produced in various substrates (Table 1). The earliest vaccines tried in humans were crude inactivated antigens prepared from Nakayama strain-infected chicken embryos (CE) or CEF or mouse brain; these were poorly or moderately immunogenic (12.5% to 77.77% seropositive), the responses being slightly higher with the CE preparation.^52,53 The dose could be reduced by one-tenth in both adults and 2–7-year-old children, and SPRs were higher (78% vs. 38%) in children, when the antigen was administered intradermally rather than subcutaneously.^53,54 The NAbs could persist for a year and be boosted,^53,55 and vaccination correlated with a gradual decrease in JE cases. .^55,56 A mouse brain-derived crude vaccine was used to immunize US army personnel in the Second World War.⁵⁷

Table 1. Landscape and other details of Japanese encephalitis vaccines licensed till date.

Type of vaccine	Virus strain and phenotype	Substrate	Generic name, if any	Manufacturer and trade name	Regimen	Country
Inactivated	Nakayama-NIH; wild-type	Mouse brain	JE-MB	BIKEN, Japan (Biken, JE-VAX); other local producers	Days 0, 7, 28 at 12–24 months of age; booster after 12–14 months, then every 3–5 yrs	European Union, India, Japan, Malaysia, North Korea, South Korea, Sri Lanka, Taiwan, Thailand, United States, Vietnam
	Beijing-1 (P-1); wild-type	Mouse brain	JE-MB, BK-VJE	GCC, Korea (JEV GCC); other local producers	Days 0, 7, 28 at 12-24 months of age; booster after 12-14 months, then every 3-5 yrs	European Union, India, Japan, Malaysia, North Korea, South Korea, Sri Lanka, Taiwan, Thailand, United States, Vietnam
		Vero cells	BK-VJE	BIKEN, Japan (JEBIK-V); KAKETSUKEN, Japan (ENCEVAC, JEIMMUGEN); Boryung Pharma, Korea (TC-JEV)	Days 0, 7, 28 at 12-24 months of age; booster after 12-14 months, then every 3-5 yrs	Japan, South Korea
		Mouse brain		China	Days 0, 7 at 12 months of age; then 2, 6 and 19 years of age	China
	Beijing-3 (P-3); wild-type	PHK cells		China		China
		Vero cells		China		China
	SA-14-14-2; live attenuated	Vero cells	JE-PIV, IC51, JE-VC	Intercell/Novartis/ Valneva, Austria (IXIARO); CSLB, Australia (JESPECT), BEL (JEEV)	Days 0, 28 as early as 2 months of age; booster after 1 yr	Australia, Bangladesh, Bhutan, Canada, European Union, Hong Kong, India, Japan, Latin America, Nepal, New Zealand, Pacific Islands, Papua New Guinea, Singapore, South Korea, Switzerland, United States
	821564XY; wild-type	Vero cells		BBIL (JenVac)	Days 0, 28 as early as 6 months of age	India
Live attenuated	SA-14-14-2; live attenuated	PHK cells	SA-14-14-2	CDIBP (CD.JEVAX)	Single dose at the age of 8–9 months; if needed, booster 3–12 months later, and at 6–7 yrs of age	China, Hong Kong, India, Japan, Nepal, South Korea, Sri Lanka, Thailand
Chimeric	YFV 17D containing JEV proteins; live-attenuated	Vero cells	ChimeriVax-JE; JE-CV	Acambis/Saofi-Pasteur (IMOJEV, THAIJEV)	A single dose for those >18 yrs, with a booster, if needed, after 5 yrs; for those ≥9 months and ≤18 years, primary dose followed by booster after 12–24 months	Australia, South Korea, Thailand

Abbreviations: BBIL, Bharat Biotech International Ltd., India; BEL, Biological Evans Ltd., India; BIKEN, The Research Foundation for Microbial Diseases of Osaka University, Japan; CDIBP, Chengdu Institute of Biological Products, China; CSLB, Commonwealth Serum Laboratories Biotherapeutics, Australia; GCC, Green Cross Corporation, Korea; i/m, intramuscular; JEV, Japanese encephalitis virus; KAKETSUKEN, Chemo-Sero-Therapeutic Research Institute, Japan; NIH, National Institutes of Health, Japan; PHK, primary hamster kidney; s/c, subcutaneous; YFV, yellow fever virus.

Note: The mouse brain- and the PHK cell-derived inactivated vaccines have been discontinued.

The earliest live attenuated virus (LAV) vaccine to be tested in humans was the OCT-541 strain, which was developed by serial passage of the wild-type strain at progressively decreasing temperatures in PHK cells. Although OCT-541 LAV showed low neurovirulence for mice and monkeys, was not infectious to C. tritaeniorhynchus mosquitoes, and failed to cause disease in a wide variety of animals, it was still lethal to suckling mice, and almost failed to elicit antibodies in both mice and cancer patients.⁵⁸

In this paper, we review the data available on the immunogenicity and protective efficacy as evaluated by randomized clinical trials, the longevity of protective responses and the effectiveness estimated through prospective studies, as well as outstanding issues relating to JE vaccines which have been licensed and used to immunise humans in various parts of the world.

Mouse brain-derived inactivated (MBDI) vaccines

For more than half a century, the MBDI vaccine was the only vaccine available for human use. The first vaccines were produced using the Nakayama strain, and the recommended regimen was a primary series of 2–3 doses at days 0 and 7–30, followed by boosters after 6–12 months, and then every 3 years.^59-61 The first trials of the crude vaccine were in Taiwan in 1965, where 80% efficacy was demonstrated.⁶² Later, with a purified vaccine, 80%-100% SPR was reported in adults from India, Japan, Thailand and the US.,^59,60,63-67 A large-scale trial (n > 65,000) involving 1-14-year-old children in Thailand showed efficacy rates of 91% with either a monovalent Nakayama or a bivalent Nakayama/Beijing-1 MBDI vaccine.⁶⁸

The Nakayama strain was replaced by the Beijing-1 (also known as P-1) strain in 1988, owing to the higher antigen yield from mouse brains, higher NAb titres in mice and humans, and better cross-neutralization and hence broader coverage of virus strains,^69-71 and the requirement for less number of doses.

The production of MBDI vaccine was discontinued in the year 2005⁷² and the last of the stocks expired in 2011.⁷³ The decision to discontinue MBDI vaccine production resulted from the observation of rare cases of acute disseminated encephalomyelitis (ADEM) in vaccinated subjects, although no causal link with vaccination could be established.⁷⁴ Additional factors led to explorations on better vaccines. One was the poor to moderate immunogenicity of the MBDI vaccine, and therefore the requirement for multiple doses as well as boosters. The other reason was the adverse events (AEs) associated with MBDI vaccine, especially in Caucasians. Hypersensitivity reactions, including rash, urticaria, angioedema, hypotension and respiratory distress were observed in subjects from Australia, Canada, Denmark, Germany, the UK, and the USA,^75-77 with higher propensity in individuals with other known allergies;⁷⁸ such AEs were also observed rarely in subjects from Japan.^76,78-80 It has been hypothesized that porcine gelatin may be responsible for the hypersensitivity reactions.^79,80 Rare cases of ADEM, encephalopathy, seizures, peripheral neuropathy, Guillian-Barre syndrome, and other neurological AEs have also been reported in Japanese and European subjects, and these AEs are thought to be a consequence of trace, undetectable levels of neuronal tissue-derived proteins contained in the MBDI vaccine.⁷⁷ Other reasons to move away from MBDI vaccines were biosafety requirements, the necessity for the use of animals, the simultaneous requirement for large number of weanling mice, the time-consuming, cumbersome and expensive purification process, the cost of the vaccine, and the aversion to vaccine-related risks vis-à-vis low JE disease burden.⁸¹

Live attenuated virus (LAV) vaccines

The only LAV vaccine that has been approved for use in humans is the SA14-14-2 vaccine, which was developed and has been extensively used in China. The vaccine strain was derived through multiple passages of the wild-type SA14 strain in animals and cell cultures.⁸² Whereas the intermediate strains retained residual neurovirulence to mice and were poorly immunogenic in mice and humans, the SA14-14-2 strain was completely attenuated for mice, and safe and immunogenic in mice, guinea pigs, pigs, and humans, including children.⁸² Compared to the wild-type SA14 virus, the SA14-14-2 LAV shows a slight growth defect (smaller plaque size, slower kinetics, lower titres) in BHK-21 cells, and is highly attenuated for neuroinvasion and neurovirulence in ICR mice.⁸³ The LAV fails to replicate in mosquitoes when fed through a blood meal, while it replicates to low titres when inoculated intrathoracically in C. tritaeniorhynchus, but with an inability to be transmitted to suckling mice,⁸² suggesting that the SA14-14-2 LAV is transmission-incompetent. The LAV strain has also been shown to be stable upon passage in mosquito or mammalian cells in vitro, or in mice (both intracerebral and intraperitoneal) or mosquitoes.⁸²

The SA14-14-2 LAV vaccine was licensed in China in 1988, and has been used there in annual spring vaccination campaigns. It is recommended for children at 1–2 years of age, with a booster one year later. Several non-randomized as well as randomized clinical trials have proved the immunogenicity and protective efficacy of the SA14-14-2 vaccine in subjects ranging from 8 months to 15 years, and being followed from 7 days to 3 years.³¹ Efficacies of 85%-95% have been reported with a single dose, or two doses separated by a short interval, in Chinese adults and children.^84,85 Two doses administered 1 or 2.5 months apart to 12-15-year-old Chinese children both produced high SPR of 94%-100%, although GMTs were higher with the latter schedule.⁸⁶ In South Korea, SPR of 96% was achieved with a single dose as against 64% with two doses of MBDI vaccine in 1-3-year-old children.⁸⁷ The schedule of the 2^nd dose being administered after a gap of 1 year has also been confirmed in South Korea.⁸⁸ In Nepal, a single dose was shown to produce an efficacy of 99.26% soon after immunization, and 98.5% one year later.^89,90 The SA14-14-2 LAV vaccine has also been shown recently to boost antibody responses when administered 1 year after a cell culture-derived inactivated (CCDI) vaccine series in children in Nepal.⁹¹

In contrast to the excellent efficacy of SA14-14-2 LAV vaccine in China, Nepal and South Korea, the efficacy in Indian adults was found to be lower as evaluated in post-marketing serosurveys: after a single dose, the 28-day SPR was 67.2% in all individuals and 73.9% in non-immune individuals, and the SPR at 1 year was 43.1% overall and only 35% in non-immune individuals,⁹² It is not clear what the reasons for the low efficacy of the SA14-14-2 LAV vaccine in the Indian populations are; the issue is yet to be resolved, and possible reasons include inaccurate reporting of coverage, possible deviation from standardized assays, absence of comparison with sera from other countries where high efficacy rates have been shown, as well as potential contribution of host genetic background.

The molecular basis of attenuation of the SA14-14-2 LAV strain has been the subject of several investigations, including those which have documented the sequential appearance of mutations during passages of SA14 wild-type to generate the SA14-14-2 LAV strain, as well as its derivative clones.^93-95 Compared to various wild-type JEV strains, up to 31 amino acid changes have been documented;⁸² one study identified 15 mutations, 1 in C and 5 in E protein of the mammalian cell culture-derived SA14-14-2,⁹⁶ whereas another, in which the viruses were cultured in C6/36 Aedes albopictus mosquito cells, identified 24 amino acid changes (1 in C, 8 in E, and the rest in other protein coding regions) in SA14-14-2.⁹⁷ No reverse mutations have been observed at all the 8 amino acids in the E protein for up to 22 passages in PHK cells, or up to 9 intracranial suckling mouse passages following 17 passages in PHK cells,⁸² suggesting that the SA14-14-2 virus is genetically stable. A seminal study showed that multiple (at least 4) mutations in the E protein are required for attenuation of neurovirulence,⁹⁸ although other studies have shown that a single mutation in the C protein (S66L), or an additional mutation in the E protein (G244E) may also lead to attenuation without loss of immunogenicity in mice.^95,99 Recent studies have also shown that the SA14-14-2 LAV is unable to produce the NS1′ protein due to the loss of the RNA pseudoknot structure as a result of a single point mutation (G66A) in the NS2A coding region,^83,100 although this does not appear to significantly affect neurovirulence.¹⁰¹

Despite the excellent safety record of SA14-14-2 LAV vaccine,¹⁰² some issues have been raised about its regulatory compliance. One issue is the consistency of cells as well as carry-over of adventitious agents, although both of these can be overcome with very strict quality control procedures, including the use of gnotobiotic hamsters. The potential for the virus to mutate during passage is another issue. A recent study showed that passage of the SA14-14-2 strain in mice, either intracranially or intraperitoneally for up to 12 passages, produced viruses which were neuroinvasive and neurovirulent.¹⁰³ In addition, concerns have been raised about the appropriateness and the methodology employed to test the lack of neurovirulence in mice.³² Furthermore, one study has reported temporal association of encephalitis following SA14-14-2 vaccination in humans, although the AE could have been due to infection with wild JEV strain since the subject was IgM⁺.¹⁰⁴ Despite these concerns, any phenotypic reversion of SA-14-14-2 LAV is probably obviated by limiting the end-of-production passage level during the manufacturing process; the stability of the critical attenuating mutations can also be ensured by deep sequencing of vaccine lots now and then. In addition, regaining virulence following multiple passages in mice is unlikely to be of any consequence to vaccination in humans since no human-to-human transmission of even wild-type JEV occurs and the SA-14-14-2 LAV is also incompetent to be transmitted by mosquitoes. Indeed, none of the aforementioned issues embody critical concerns, and the compliance of the manufacturing process to international standards has led to the pre-qualification of the SA-14-14-2 LAV vaccine by the World Health Organisation.

However, the SA-14-14-2 LAV may present a threat to swine, as evidenced by the isolation of very closely related virus strains obtained from aborted fetuses and stillborn piglets in a herd which had been vaccinated with the SA-14-14-2 LAV vaccine. Incidentally, the isolates were also fatal to suckling mice.¹⁰⁵ Whether amplification of the SA-14-14-2 LAV in pigs, and reversion to virulence, if any, are risks that pose the potential for the vaccine virus to be transmitted to humans through mosquitoes, and further cause any disease, is an open question that needs to be addressed. In this context, another LAV vaccine, ML-17, which was derived from the wild-type strain JaOH0566, by passaging serially in monkey kidney cells under descending series of temperature,¹⁰⁶ is the only LAV vaccine approved for use in swine in Japan, and has been found to be attenuated in mice.¹⁰⁷ The fact that pigs act as amplifying hosts, and in the absence of extensive data showing that LAV vaccines are genetically stable and produce low, mosquito transmission-incompetent level of viremia in pigs, such vaccines should not be used in pigs to minimise the risks of reversion.

There have been other attempts to develop LAV strains for use as vaccines. A study reported that serial passage of four wild type strains in cell culture resulted in two of them being attenuated for mice.¹⁰⁸ In another study, three of the six infectious clones generated from the JaOArS982 wild-type strain produced recombinant viruses which exhibited attenuated phenotype, with no viremia or neuroinvasion in mice; one recombinant virus was shown to be protective against challenge.¹⁰⁹ Protective immunity, NAbs, splenocyte proliferation, and lymphocyte-mediated cytotoxicity could be demonstrated with another attenuated strain.¹¹⁰ However, none of these progressed beyond animal studies.

Genetically engineered vaccines

Two genetically engineered vaccines have been tested for immunogenicity in humans. One is the recombinant poxvirus vaccine, where the prM and E protein genes of JEV were expressed through vaccinia or canarypox viruses. Two doses of these vaccines could elicit NAbs in monkeys,¹¹¹ and both antibody and T-cell responses in a small number of human subjects.^47,108 However, these vaccines were less immunogenic as compared by GMTs, and more reactogenic compared to the MBDI vaccine,¹¹² and were not developed further.

The only genetically engineered licensed vaccine is JE-CV, previously known as ChimeriVax-JE, in which the prM and E genes of yellow fever virus (YFV) 17D vaccine strain have been replaced by the respective genes of JEV. JE-CV was designed to be attenuated; residual neurovirulence was observed when genes from Nakayama strain, and not SA14-14-2 LAV strain, were used,¹¹³ and hence the latter was used for further development. The genotype as well as the attenuated phenotype was found to be stable for at least 18 passages in cell culture and 6 passages in the mouse brain.¹¹⁴ The virus could only produce transient, low level of viremia upon intracranial inoculation of non-human primates, with no virus excretion.^115,116 Further, doses ranging from 3.0 to 5.0 log₁₀ plaque-forming units (PFU) could protect monkeys.¹¹⁷ As far as genotypic characterization of JE-CV is concerned, no single amino acid in the E protein appears to be important for neurovirulence, as evaluated in mice, and at least four residues have been shown to be necessary for neurovirulence in a high proportion of animals.^98,118 Although another study showed that a single mutation (M279K in the E protein) produced when the virus was propagated in foetal monkey liver cells could revert the non-neurovirulent chimera to a neurovirulent virus as evaluated by 50% lethal dose and survival distribution in suckling mice and by brain histopathology in rhesus monkeys, the derived virus was still less neurovirulent to monkeys than the YF17D vaccine virus.¹¹⁹ In addition, these results do not apply to the marketed vaccine, which is genetically stable under manufacturing conditions. Moreover, JE-CV produces low level of viremia (1.0–1.5 log₁₀ PFU/mL of serum) in humans,¹²⁰ and replicates restrictively in Culex and Aedes mosquitoes,¹²¹ factors which together probably lead to the transmission-incompetence of JE-CV.

NAb responses to a single dose of JE-CV vaccine were as good as that elicited by 3 doses of MBDI vaccine in adults,¹²² and a second JE-CV dose could induce memory responses even after 9 months, although the booster effect was modest.¹²⁰ In Thai and Philippine children aged 12–24 months, the vaccine could produce 96% SPR after a single dose,²²¹²³ and boost NAb responses elicited earlier by MBDI vaccine with SCR of >90% and SPR of 100% in 2-5-year-old children,³¹ as well as 2 years following priming with JE-CV.¹²⁴ In a head-to-head comparison of JE-CV and SA14-14-2 LAV vaccines in 12-18-month-old Thai and 12-24-month-old South Korean children, SPRs were equivalent to each other, whereas 28-day GMTs were higher, albeit not statistically significant, with JE-CV than with SA14-14-2 LAV vaccine.,^125,126 JE-CV has also been shown to elicit anamnestic responses and significantly increase GMTs in 1-to-5-year-old Thai children who had been vaccinated 9–13 or 12–24 months earlier with the SA-14-14-2 LAV vaccine;^127,128 incidentally, the responses were slightly, but not significantly higher when the booster with JE-CV was later than earlier.¹²⁷

Cell culture-derived inactivated (CCDI) vaccines

Cell culture-derived inactivated (CCDI) vaccines have now replaced the animal tissue-derived vaccines, owing to a more controlled production process of the former. The first such vaccine was the PHK cell propagated, formalin-inactivated wild type OCT-541 vaccine; in animal studies, this vaccine showed 2–3 times higher potency than the OCT-541 LAV vaccine.¹²⁹ The inactivated vaccine elicited NAb responses in 92% of the subjects after a primary series of immunization; in fact, SPRs of 85% and 62% were observed even after 4.5 and 8 months, respectively, and a booster at 8 months elicited protective titres in 100% of the individuals, with higher NAb titres.¹³⁰ However, this vaccine has not been used for routine or campaign immunizations. Another vaccine produced using the wild-type Beijing-3 (P-3) strain in PHK cells, was used in humans for some time only in China, and has since been discontinued.

The first methodically developed CCDI vaccine was the formalin-inactivated vaccine from the Vero cell-adapted SA-14-14-2 LAV strain. In clinical trials, both GMTs and SCRs produced by this vaccine (IC51) at a dose level of 6–12 µg were higher than that produced by the MBDI vaccine.¹³¹ It was later shown to produce SCR comparable to or higher than that by the MBDI vaccine, but with higher GMTs in subjects from Australia, Germany, northern Ireland, and the US; the SCRs were 95%-100% for up to 6 months after vaccination.^132,133 It was also shown that two administrations of a standard dose (6 µg) was better than single administration of a higher dose (12 µg) in adults.¹³⁴ In 1-3-year-old children, this vaccine produced SCRs of 65%-71%, with no significant difference between 3 or 6 µg per dose, after one dose, and 95% after two doses, being equivalent to two or three doses, respectively, of the MBDI vaccine.¹³⁵ The vaccine could also boost responses earlier elicited by MBDI vaccine in Finnish and Swedish travelers,¹³⁶ as well as in US marines.¹³⁷

Another CCDI vaccine, but derived from the P-1 strain,¹³⁸ was demonstrated to be as immunogenic as the MBDI vaccine in a Phase I study in adults; SCR was 96.7% with this vaccine as compared to 92.9% with the MBDI vaccine, and GMTs were similar between the two vaccines after the first and the second dose, but higher for the CCDI vaccine after the third dose.¹³⁹ A single dose may be sufficient for some age groups, as observed in Japanese adults aged 25–39 years, but two doses have been shown to be required for the best results.¹⁴⁰ In later Phase III studies involving 6-90-month-old children, >94% SCR could be achieved with two doses of 5 or 2.5 µg,¹⁴¹¹⁴² and with three doses of 1.25 µg.¹⁴² Studies comparing this CCDI vaccine to MBDI vaccine in Korean 12-23-month-old children confirmed the non-inferiority of the former to the latter in SCRs, although GMTs were much higher with the former, especially after 2 or 3 doses.¹⁴³ A booster dose at 2–4 years following primary series at 1–2 years could dramatically increase the GMTs.¹⁴⁴

Another CCDI vaccine based on Vero cell-adapted strain, designated 821564XY, from India has also been reported recently. Immunogenicity studies of 2 doses of this vaccine compared to a single dose of SA14-14-2 LAV vaccine showed that the former was better in all age groups ranging from 1 to 50 years, with SPRs of >90% after both the doses (immunized on days 0 and 28, and sera obtained at days 28 and 56), whereas SPRs for SA14-14-2 LAV vaccine were 77.6% and 60.3% at the same time-points following a single dose.²⁹ Although this difference was probably due to a combination of the single dose of SA14-14-2 vs. the 2 doses of the CCDI vaccine and the neutralization data being obtained with the strain used for the CCDI vaccine, it was found that the SRC, SRP and GMT were all higher for the homologous virus-sera pairs for the CCDI vaccine strain than the SA-14-14-2 LAV strain.²⁹

It has been noted that, compared to MBDI vaccines, the Vero cell-derived inactivated vaccine elicits markedly higher NAb responses, or that lower amounts of antigen are sufficient to elicit equivalent responses.^{74,132,141,142,145} Although the factors responsible for this are not known, some of the suggested reasons for higher potency of CCDI vaccines are: (a) less bioprocess steps and milder down-stream processing conditions, (b) ultrastructural morphology of cell culture-derived vaccine virus being more similar than the MBDI vaccine virus to the native virus, (c) higher affinity toward several monoclonal antibodies,¹⁴¹ and (d) differential glycosylation of the E protein.¹⁴⁶

Cross-reactivity and cross-protection of antibody responses to JE vaccines

Immunological cross-reactivity refers to responses which are not restricted to a single definable type of a pathogen. Whereas cross-reactivity is often common among related pathogens or different types of the same pathogen, cross-neutralization by sera, which defines specific viruses and serotypes, is less common. Although JEV exists as a single serotype, questions have been raised about the specificity and reactivity of vaccine-elicited immune responses to various genotypes. Several antigenic variants of JEV have been reported as evidenced by the reactivity of monoclonal antibodies to the E protein, or in haemagglutination inhibition (HI) or neutralization assays.^147-154 A recent study reported complete absence of reactivity of sera to heterologous genotype virus when guinea pigs were immunized with G-I or G-III viruses.¹⁴⁹ However, it is difficult to interpret these results without performing PRN assays; indeed, contradicting results were obtained between HI and neutralization assays by the latter researchers.¹⁵⁵ Concerns about effectiveness of vaccines against the various genotypes have been compounded by the fact that G-III, which was the predominant genotype circulating up until a decade ago,¹⁵⁶ and hence has been the only genotype chosen to produce all the vaccines, has now been dominated by viruses belonging to G-I in endemic areas.^{23,40,157-162}

Several studies have evaluated the ability of sera obtained from clinical trials to neutralize viruses belonging to different genotypes as well as different strains belonging to G-III. An early study on the inactivated vaccine in Japan reported SCRs between 85% and 100% at 1–3 months after two doses, between 67% and 92% one year later, and 100% after a booster, against nine different JEV strains of unknown genotype isolated in China, Japan and Thailand over almost 50 years.¹⁶³ A similar observation was made against three different derivatives of Nakayama strain as well as three other strains of unknown genotype.⁵⁹ In another study from Taiwan, SPRs were higher against Nakayama than a local wild-type isolate; in addition, antibodies to the wild-type isolate could neutralize the Nakayama strain with all the subjects but only 37.93% of the subjects with antibodies to the Nakayama strain could neutralize the wild-type strain.¹⁶⁴

In several studies, immunogenicity (SCR, GMT) of vaccines were observed to be slightly higher for the homologous vs. heterologous G-III strains (e.g. Nakayama vs. Beijing-1 or SA14-14-2, or any of these vs. other wild-type G-III strains) with MBDI, SA14-14-2 LAV, JE-CV as well as Vero CCDI vaccines, although the differences in SCRs or GMTs are not significant statistically.^{22,24,25,27,31,65,70,71,117,120,122,136,165-167} This differential neutralization as well as protection against G-III viruses has also been shown in mouse challenge studies.¹⁶⁸ Other studies have shown no such differences or higher GMTs with the heterologous virus strains.^31,122 As expected from the nature of measurements, most studies have reported differences in GMTs and not SPRs.

A few studies have examined cross-neutralization of sera against viruses belonging to different genotypes. In general, these studies have shown that NAb titres are highest against G-II and homologous G-III viruses, followed by G-IV and heterologous G-III viruses, whereas titres against G-I viruses are the lowest, and in some cases below protective levels.^22,24-29 A recent study specifically evaluated the ability of a G-III virus to elicit protective responses in mice, and the ability of human post-vaccinal and acute phase sera to neutralize G-V viruses. The study found that (a) NAb titres in mice immunized with either G-III or G-V virus were higher with homologous viruses but poor against G-V, (b) the SA14-14-2 LAV or the P-3 inactivated vaccines were less protective against G-V virus challenge in mice, and (c) both GMTs and SPRs were poor against G-V virus, intermediate against G-I virus, and highest against G-III virus with both post-vaccinal and acute phase human sera.⁴³ However, (a) although compared in the same assay, this study used the 90% PRNT rather than the standard PRNT₅₀, (b) it is not clear which genotype virus had infected the patients who donated the acute phase sera, and (c) why the cross-reactivity is higher with sera obtained from older patients.

Although titres against homologous viruses were found to be generally, but not always, higher than against heterologous viruses, if and how differences in cross-neutralization translate to differences in clinical protection is unknown. It is important to note that, in most cases, SPRs are not significantly different, suggesting no consequence of the differences in GMTs to clinical protection. Therefore, vaccination of humans, which are dead-end hosts for JEV, would be predicted to have no differential impact on the natural transmission cycle and the prevalence of different genotypes. However, it remains to be explored whether differential cross-neutralization measured typically at the height of antibody response translates to similar differences in neutralization of different genotypes and whether this would impact long-term protection. One study evaluated NAb responses in Thai children at 12–18 months of age with MBDI, followed by a booster with JE-CV 6–39 (average 12) months later. The GMTs were highest against G-I and G-III viruses, followed by G-II and then by G-IV viruses prior to booster, and at each of 28 day, 6 month and 5 year time-points post-booster, the GMTs followed the same pattern of G-III > G-I and G-II > G-IV. The SPRs were 87.2% against G-II and G-III viruses but 78.7% against a G-IV virus prior to booster, but were within a narrow range of 91.7%-97.9% when boosted.¹⁶⁹ These studies indicate that vaccines carrying G-III virus antigen can elicit similar long-term SPRs against all the major genotypes, but further studies are needed to address this issue with vaccines delivered as a single dose as well as when administered after a gap of 5 or more years. The results of such studies also need to be correlated with protection against disease caused by different genotypes in the field, a proposition which is very difficult to investigate because of lack of co-circulation of the various genotypes in the field at the same time. In any case, a thorough investigation involving a wide array of characterized strains belonging to each of the genotypes is warranted to conclude on the ability of the currently available vaccines to generate protective NAbs against the various circulating JEV strains. Furthermore, this exercise would contribute to the development of reference viruses and sera as well as globally standardized assay procedures to evaluate immunogenicity of JE vaccines.

Longevity of protective responses induced by JE vaccines

The longevity of protective responses following vaccination has been investigated by several groups. Seroprotective titres following immunization with 2–3 primary series of MBDI vaccine appear to last 1–3 years, and can then be boosted geometrically. In the initial studies in Thailand and Japan, SPRs of 60%-88% were observed after one year,^59,63-65 whereas SPRs waned in 33% of the subjects by 6 months in US soldiers, but could be boosted to 100% soon afterwards;⁶⁰ on the other hand, 32% SPR was observed even after 3 years in Torres Strait, Australia.¹⁷⁰ Following 3 doses, protective titres have been observed for at least 3–5 years.⁷¹ By contrast, in India, immune response to an indigenous preparation using the Nakayama strain was shown to be around 50%, with retention of SPRs of around 35% at 13–17 months.¹⁷¹

Based on mathematical modeling using data following a two-dose MBDI vaccine, it was estimated that seroprotective titres may last as long as 5, 10, or 15 years in 82%, 53%, and 18% of the subjects, respectively.¹⁷² In a second modeling study, it was estimated that it would take 805 and 355 days to reach SPRs of 50%, when evaluated by PRNT₅₀ and PRNT₈₀, respectively.¹⁷³ CD4⁺ T-lymphocyte responses elicited by MBDI vaccine were also detected several months after the third immunization,⁴⁸ although the contribution of this response to prevention of disease or infection is not clear.

The longevity of protective antibody responses induced by SA14-14-2 LAV vaccine as well as the JE-CV vaccine have been the subject of some studies which have been conducted in both endemic (China, Nepal, South Korea, Thailand) as well as non-endemic (Australia) countries. In Nepal, a single dose of the SA-14-14-2 LAV vaccine resulted in the persistence of protective NAbs for 1 year in 98.5%, for 4 years in 89.9%, and for 5 years in 63.8% of the individuals.¹⁷⁴ In Thai children immunized with a single dose at the age of 12–24months, 87% and 84% SPRs were observed at 7 and 12 months, respectively, whereas in children previously immunized with the MBDI vaccine and then boosted at 2–5 years of age with a single dose of JE-CV vaccine, the respective proportions were 100% and 97%.²² A very high level of boosting was also achieved in Thai children immunized at 12–18 months with the MBDI vaccine, followed by booster 6–39 months later with the JE-CV vaccine; SPR was 87.2% prior to booster, and >95% even after 5 years following the JE-CV booster. The GMTs were well above protective levels and 2-3-fold higher at 5 years post-JE-CV booster than prior to JE-CV administration.¹⁶⁹ In Australian adults, a single dose of JE-CV vaccine resulted in persistence of protective NAbs in 95%, 90% and 93% of the subjects at 12, 24 and 60 months, respectively, and in 99%, 99% and 97% of the subjects at the respective times after a booster at 6 months.¹⁶⁶ Modeling study based on responses in adults living in endemic countries observed up to 5 years following two immunizing doses separated by 6 months predicted a gradual slow decrease in SPR from 93.4% to 85.5% from 6^th to 10^th year.¹⁷⁵

In subjects immunized with a CCDI vaccine (IC51), SCRs of >91% were observed at each of the 6, 12, 18 and 24 months.^131,176 SPRs following a 2-dose series were 82.8%, 58.3% and 48.3% at 6, 12 and 24 months after the last dose, and the GMTs could be significantly increased, and be maintained at high titres for at least a year, with a booster dose after 11 or 23 months.¹⁷⁷ In a modeling study based on responses for up to 12 months following a booster at 15 months, it was estimated that protective NAbs would remain for 3.8, 6.9, and 8.0–9.0 years if the GMTs at 30 days post-booster were 100, 500, and 900–1500, respectively.¹⁷⁸ The durability of the response predicted by modeling studies was confirmed when responses in that study were evaluated 76 months after the 15-month booster, when the SPR was 96%; based on the earlier and these findings, the estimated longevity of protective responses were revised to 7.4, 12.5, 14.4–14.8, and 16.1 years if GMTs were 100, 500, 900–1000 and 1500, respectively, with an average protection of 14 years.¹⁷⁹ With an Indian CCDI vaccine, it was shown that about two-thirds of the evaluated subjects carried protective NAb titres for up to at least 18 months.²⁹

The results on the longevity of responses should be interpreted with caution since the outcome may depend on whether the study was conducted in a non-endemic or an endemic area. Seropositivity rates are known to increase significantly beyond 10–15 years of age in endemic areas,^167,180-182 and an inverse correlation has been observed between an age of >5–6 years and the incidence of JE disease.¹⁶⁷ In addition, direct correlation between seropositivity rates and the incidence of JE in an area has also been documented.¹⁸³ Furthermore, asymptomatic infections with other flaviviruses in endemic areas is expected to boost responses elicited by JE vaccination. It has been shown that previous immunization against tick-borne encephalitis (TBE) or YF can enhance NAb responses to inactivated JE vaccine,^116,133 while other studies have shown that pre-existing immunity (PEI) to flaviviruses, including JEV, dengue virus, TBEV or YFV do not appear to influence the NAb responses to JE vaccine.^{22,27,120,123,124,170,179,184,185} On the other hand, prior immunization with a CCDI JE vaccine was found to elicit cross-reactive antibodies which enhanced immunogenicity of the YF vaccine.¹⁸⁶ In essence, the duration of protective responses is expected to be longer in endemic countries where there is natural exposure to flaviviruses, as well as in those who have been previously immunized with related flavivirus vaccines, and relatively shorter in completely naïve individuals in non-endemic regions.

Effectiveness of JE vaccines and the benefit in reducing the burden of disease

Effectiveness of a vaccine is estimated in non-ideal, but real-life settings, with few to no exclusion criteria, and is different from efficacy, which is enumerated from controlled clinical trials. It is difficult to carry out prospective randomized clinical trials to evaluate the effectiveness of JE vaccines because of (a) the requirement for a large sample size (to compensate for the low attack rates in the field), and (b) the impossibility of maintaining a placebo control for a disease with significant mortality and sequalae. Therefore, effectiveness of JE vaccines has been extrapolated from a few retrospective as well as prospective case control studies.

A 30-year retrospective study of vaccination and incidence records in Taiwan concluded that the MBDI vaccine had an estimated effectiveness of 96.98% against disease incidence, 97.57% against disease mortality, and 19.3% in decreasing the fatality of confirmed cases in 1-14-year-old children.¹⁸⁷ The estimated effectiveness was 85.59% after one dose, 91.07% after two doses, and 98.51% after three doses.¹⁸⁷ In children vaccinated at 1-year of age in Viet Nam, the estimated effectiveness of a 2-dose primary series followed by a 1-year booster of MBDI vaccine was 92.9%.¹⁸⁸ In Thailand, the effectiveness was found to be 100% and 94.6% for children of less than and more than 18 months of age, respectively, and 84.8% for all ages between 1 and 6 years; also, the effectiveness was similar between two and three doses.¹⁸⁹ In a rural province in China, the effectiveness of SA14-14-2 LAV vaccine was determined to be 80% with a single dose and 97.5% with 2 doses.⁸¹ In Nepal, estimated effectiveness of >98% could be achieved as early as 2 weeks after vaccination of 1-15-year-old children,⁸⁹⁹⁰ and remained high (96.2%) even after 5 years following a single dose.¹⁹⁰ In India, one case-control study showed that the SA14-14-2 LAV vaccine had an effectiveness of 94.5% for up to 6 months after vaccine administration,¹⁹¹ whereas another unmatched case-control study showed that the vaccine had an effectiveness of only 79%.¹⁹² It is important to note in interpreting these results that, as with longevity of immune responses, the effectiveness can be influenced by endemicity or non-endemicity of JEV and/or other flaviviruses, and whether or not there was previous exposure to other flavivirus vaccines.

The real reflection of the effectiveness of a vaccine is the reduction in the number of clinical cases. More than several million cases of JE were reported from East Asia before the widespread use of vaccines; as much as 2% of the Chinese population has been reported to be affected between 1966 and 1971.¹⁹³ Vaccination for JE has been shown to decrease (a) the incidence of both JE and acute encephalitis syndrome (AES) in several countries or regions, including China,,^194,195 India,¹⁹⁶ Japan,^197,198 Malaysia,¹⁹⁹ Nepal,^200,201 South Korea,⁷⁵ Sri Lanka,²⁰² Taiwan,²⁰³²⁰⁴ and Thailand.⁴⁵ The introduction of JE vaccine reduced morbidity in China by 97% from 1971 to 2005.¹⁹⁴ In Nepal, a 72% reduction in JE, and a 58% reduction in AES cases were observed in a short span of five years, from 2004 to 2009;²⁰¹ and an estimated reduction in JE cases by 84% and deaths by 92% has also been predicted.²⁰⁵ The risk of acquiring JE was estimated to have been reduced by 61% in Sarawak, Malaysia;¹⁹⁹ the reduction was significant even when controlled for environmental variables in Nepal and Sarawak.^199,206 In Thailand, the proportion of encephalitis caused by JEV had reduced from 40% to about 15% by the early 2000s,⁴⁵ and childhood vaccination has been associated with an increase in the age at JE attack from 3–6 years to >9 years in Malaysia.²⁰⁷

Several studies have explored the benefit of JE vaccination in terms of saving lives, reducing treatment costs and disability care, and saving future earnings. Analyses from Thailand have predicted that savings per case would be around US$ 70,000, and that vaccination at 18-months of age would be cost-effective unless the incidence of JE remained below 3 per 100,000 in unvaccinated population.²⁰⁸ Models have predicted that vaccination could prevent more than 400 JE cases and more than 100 deaths due to JE, and save around 6500 disability-adjusted life years (DALY) and US$ 0.3–0.5 million per 100,000 persons in Shanghai; whereas prevention of JE cases or deaths, or savings were similar between the inactivated (P-3 strain) and the SA-14-14-2 LAV vaccines, the savings were estimated to be 47% greater for the latter than the former vaccine.²⁰⁹ Immunization of 100,000 persons in Guizhou province in China has been predicted to save US$ 1.6 billion for the health system, and US$ 11.6 billion for the society.²¹⁰ Modeling of the cost-effectiveness of JE vaccination in Bali, Indonesia, has been predicted to save US$ 700 per case and US$ 31 per DALY.²¹¹ In Cambodia, vaccination has been estimated to avert up to 2888 JE cases, 376 deaths and 2354 disabilities, potentially saving US$ 21.84, 52.85 or 34.89 per DALY if routine, campaign, or combined immunizations, respectively, were to be carried out, consequently saving US$ 1.6 million in total.²¹² A study from a highly endemic part of India has estimated that investment in JE vaccination could yield 11-fold benefit per person over five years, and provide a discounted net benefit of US$ ∼10 million.²¹³ A more comprehensive analysis for 14 endemic countries for the period of 2007 to 2021 estimated that campaign and routine immunizations would result in a decrease of 193,676 cases, 43,446 deaths, and 77,470 cases with sequelae, reducing 6,622,932 DALYs, and saving about US$ 19 million in acute case hospitalization costs.²¹⁴ Thus JE vaccines are effective and highly beneficial.

Summary and outstanding issues

Japanese encephalitis (JE) is a vaccine-preventable disease, and vaccination is the crux of our efforts to control JE. Inactivated, live-attenuated as well as chimeric vaccines have been licensed for human use, and all have been shown to be efficacious against more than one strain and genotype. Importantly, deployment of JE vaccines, whether on routine or on campaign basis, has effectively reduced the burden of disease to a great extent in endemic countries. Furthermore, all the vaccines have also been shown to induce fairly durable protective responses following recommended schedules. However, the following critical and outstanding issues need to be addressed in controlling JE as well as in further understanding immune responses to and effectiveness of JE vaccines:

1. Vaccination needs to be implemented under the universal/national immunization program in all the endemic countries, and the importance of high percentage coverage cannot be emphasized enough. With continuous vaccination through mandatory childhood immunization, as well as concerted international efforts, including vaccination of domestic pigs and vector control, it may be possible to control, if not eliminate JE globally.

2. Issues on standardized reagents and procedures to evaluate immunogenicity and protective ability of the vaccines need to be resolved. It may be necessary to identify several virus strains and reagents for each of the genotypes. On the other hand, specific cell systems need to be mandated for neutralization assays. In addition, potency tests based on animal challenge studies may need to be developed.

3. Head-to-head comparison of the vaccines, especially in pediatric subjects, is required. It is difficult to carry out a systematic analysis of the results obtained from the various clinical trials for the different vaccines owing to differences in vaccine types, doses and schedules, variations in study designs, population and sample size, the level of endemicity, and the evaluation of end-points. A recent post-hoc analysis concluded that both the SA-14-14-2 LAV and a Vero cell-derived CCDI vaccine were equivalent to each other, and better than the PHK-derived CCDI vaccine, in terms of SCR.²¹⁵ However, it may be important to compare GMTs, since these may predict the durability of responses. It is also important to compare the AEs to different vaccines in pediatric subjects in head-to-head trials.

4. Although effectiveness studies are difficult if not impossible to carry out, more case-controlled retrospective studies based on vaccination and epidemiological records are required to better estimate the effectiveness of JE vaccines.

5. Further investigations are needed on potential interference with or enhancement of responses in individuals with pre-existing anti-flaviviral immune responses. It may also be necessary to develop more specific diagnostics to facilitate serological diagnosis of encephalitis in subjects if previously vaccinated for JE.

6. Studies are also required to evaluate interference, if any, with other childhood vaccinations so that JE vaccine can be administered simultaneously with pediatric vaccines, thereby increasing compliance.

7. The elicitation of CMI responses and their long-term duration need to be evaluated. It is likely that longevity of protective antibody responses will be dictated by CMI responses, and it is possible that the overall protection against JE is a combination of the effect of antibodies and CMI responses. In this context, one major question is whether NAbs need to be present, and at what levels at the time of infection, or whether an anamnestic B and/or T cell responses is sufficient for the maintenance of protection against JE. Animal models, including those of humanized mice, in combination with adoptive transfer of immune sera, or cells, can be employed to answer these questions, but the results still might not be applicable to situation in humans. One way to understand the contribution of CMI to protection against JE would be to design a longitudinal study to evaluate the quality, magnitude and breadth of immune responses over several years and correlating cellular responses, NAb levels and disease outcomes. However, such studies may be difficult to carry out in highly endemic areas where natural challenge can contribute to maintenance of responses; hence, the studies need to be performed in a carefully selected area of low to intermediate endemicity. An interesting possibility is to follow only vaccinated individuals by the absence of antibodies to NS1′ protein, since such antibodies are only elicited by replicating wild-type JEV strains,¹⁰¹ as well as other related flaviviruses.^{216, 217}

8. Alternate means of vaccine administration, e.g., mucosal or dermal delivery, need to be evaluated in clinical trials. These could improve compliance with immunization regimen, especially in pediatric populations. These approaches may also provide dose-sparing, which is critical in mass immunization.

9. There is also a need to explore the contribution of immunizing pigs during certain seasons. It is well known that appearance of virus in pigs predates circulation of the virus among humans, and prior vaccination of pigs in targeted areas may reduce virus transmission. Whereas immunizing domestic pigs is possible, novel strategies such as oral bait vaccines may need to be designed to immunize feral or nomadic pigs. However, constant vigilance will need to be exercised to cover all susceptible pigs due to the fact that pigs farrow more than once a year and produce a litter size of 4–6 on an average.

Abbreviations

ADEM	=	acute disseminated encephalomyelitis
AE	=	adverse event
AES	=	acute encephalitis syndrome
BHK	=	baby hamster kidney
CCDI	=	cell culture-derived inactivated
CE(F)	=	chicken embryo (fibroblast)
CMI	=	cell-mediated immunity
DALY	=	disability-adjusted life year
E	=	envelope
G	=	genotype (-I, - II, -III, -IV, -V)
GMT	=	geometric mean titer
IFN	=	interferon
JE(V)	=	Japanese encephalitis (virus)
JE-CV	=	JE chimeric vaccine
LAV	=	live attenuated virus
MBDI	=	mouse brain-derived inactivated
NAb	=	neutralizing antibody
NS	=	non-structural
PEI	=	pre-existing immunity
PFU	=	plaque-forming unit
PHK	=	primary hamster kidney
prM	=	pre-membrane
PRN (T50)	=	plaque reduction neutralization (titer-50%)
SCR	=	seroconversion rate
SRP	=	seroprotection rate
TBE(V)	=	tick-borne encephalitis (virus)
YF(V)	=	yellow fever (virus)

Disclosure of potential conflicts of interest

Ella Foundation, to which NRH belongs, was involved in clinical evaluation, data analysis and publication relating to JENVAC, a JE vaccine of Bharat Biotech International Limited (BBIL), Hyderabad, India. MMG advised BBIL during the development of the same vaccine. Ella Foundation also advises BBIL on production and assay development for their other products. The authors have no other financial or personal conflicts of interest.

Acknowledgments

We thank the anonymous reviewers who provided excellent insights and contextual interpretations which substantially improved the manuscript.